7 Dangerous Myths DBAs Believe about Data Modeling
4 likes•1,352 views
The document discusses seven dangerous myths about data modeling, emphasizing their misconceptions and the importance of enterprise data modeling (EDM) in enhancing project outcomes. It highlights the value of data models beyond mere documentation, arguing that they can improve design, reduce errors, and increase agility in response to business needs. Additionally, it addresses common beliefs like the ineffectiveness of generated DDL and the notion that traditional ORM scripts suffice for data modeling.
7 Dangerous Myths DBAs Believe about Data Modeling
- 1. 7 Dangerous Myths DBAs Believe about Data Modeling Karen Lopez and John Sterrett
- 2. Yes, Please do Tweet/Share today’s event @datachick @JohnSterrett @ERStudio
- 3. Karen López Karen has 20+ years of data and information architecture experience on large, multi-project programs. She is a frequent speaker on data modeling, data-driven methodologies and pattern data models. She blogs at community.embarcadero.com, datamodel.com and Dataversity.net She wants you to love your data.
- 4. John Sterrett John has 10+ years of experience with SQL Server performance tuning, high availability, disaster recovery, database design, development and database administration. He is a frequent speaker on performance tuning, high availability, disaster recovery and proactive monitoring. He blogs at community.embarcadero.com , johnsterrett.com , and sqlauthority.com. He likes to use tools and automate processes so he has more time playing with his family and mixing records on his turntables.
- 5. About You... Who Are You?
- 7. Outcomes TODAY Understand & Describe Enterprise Data Modeling Process and Values Describe the Costs, Benefits & Risks of EDM-driven projects Understand how Modern Data Modeling efforts contribute to agility Understand how Modern Data Modeling efforts contribute to quality
- 8. 7 Dangerous Myths 1. Data models are only for creating new databases 2. Data modeling is just documentation 3. DDL generated from data models is unusable 4. Data modeling isn’t needed – ORM scripts are good enough 5. I can code that myself faster & better than the data modeling tool can 6. All normalization just slows things down 7. Data modeling is just boxes and lines (the diagram is the model)
- 9. What Other Myths Do You Hear? Put them in the Q&A!
- 10. Dangerous, really? Computers are good at syntax and consistency, humans aren’t A non-trivial database will have tens or hundreds of thousands of objects in it Human error costs money, delays projects and puts business and customers at risk
- 11. Facts of Life, 2015 Faster staff turnover on projects Geographically distributed resources Overloaded with projects IT must be agile Just the right “documentation” Models are better than documents Modeling leads to better designs Just go ask the guy who wrote it.
- 13. Enterprise Data Modeling encourages a practical balance between enterprise and project points of view comprises both application and enterprise data models enables IT groups to respond more quickly and effectively to business needs delivers information that is the most useful to the business uses the proper tools and techniques in delivering project outcomes
- 15. More than New Databases Reverse Engineering More than a picture Vendor Applications Legacy Databases Data Lineage
- 22. Package Application Project Logical Modeling 22
- 23. ORM scripts …are good enough
- 24. ORM Code… SELECT "ORMTable"."Name" ,"ORMTable"."Address1" ,"ORMTable"."City" ,"ORMTable"."State" ,"ORMTable"."ZipCode","ORMTable"."PKID" ,"ORMTable"."InceptionDate" ,"ORMTable"."PostDate" ,"ORMTable"."Address2" ,"ORMTable"."LoanID" FROM "Database"."dbo"."ORMTable" "ORMTable" INNER JOIN "Database"."dbo"."ORMTable" "ORMTable2" ON ( ("ORMTable"."PKID" = "ORMTable2"."PKID") AND ("ORMTable"."InceptionDate" = "ORMTable2"."InceptionDate") ) AND ("ORMTable"."PostDate" = "ORMTable2"."PostDate") WHERE ( "ORMTable"."PKID" = N'ABC1234' AND ( "ORMTable"."InceptionDate" >= {ts '2015-06-07 00:00:00' } AND "ORMTable"."InceptionDate" < {ts '2015-06-07 00:00:01' } ) AND ( "ORMTable"."PostDate" >= {ts '2015-04-08 00:41:26' } AND "ORMTable"."PostDate" < {ts '2015-04-08 00:41:27' } ) AND "ORMTable"."ID" = N'1' OR "ORMTable2"."PKID" = N'ABC1234' SELECT * Multiple pages of SQL for a simple query Full Table Scans
- 25. Unusable DDL/DML
- 26. Really?? Once upon a time, you might be able to justify that Data Modeling tools didn’t develop usable DDL and DML scripts. This ship has sailed.
- 27. Don’t break my source code… I spent hours developing my code. Please don’t break it! SELECT * FROM TABLE SELECT * FROM TABLE ORDER BY 2 DESC
- 28. I can script it better & faster
- 29. Data Models are Complex ….
- 30. Requirements Data Model Database* More requirements / changes / tuning / whims + Non Model Stuff Data Model Driven Data Model Driven
- 31. Just boxes and lines
- 33. Metadata • More information than goes into the database • Used to manage models • Can be used to manage • Security Requirements • Privacy Requirements • Stewardship • Quality Requirements • Semantics of data
- 36. Is all about the keys, ‘bout the keys, ‘bout the keys… Depends on understanding the MEANING of the keys and columns Goes all to heck* when you have surrogate keys Depends on the make up of the key parts (columns) Normalization
- 37. Normalization, briefly 1NF – all instances (rows) have the same facts (columns). There are no repeating duplicate columns 2NF – only applies to multi-part keys. No fact is about just part of the key 3NF – No fact depends on another non-key column
- 38. 3NF Every fact is either part of a key or depends upon the key, the whole key, and nothing but the key. ….so help you Codd Michael J Swart
- 40. Remember… “Normalization is like marriage… …you always end up with more relations” -Data Modeling Essentials, 3rd Edition Simsion & Witt
- 41. It’s taught wrong? Taught as a process, not a measurement 1NF, 2NF, 3NF, etc. Used like a grade, instead of a measurement Magical “3NF”
- 42. No data professional Builds a data model/design by going from 1NF to 2NF to 3NF to …. Says a database has a normal form (only tables have normal forms) Thinks of #NF as a “Grade” for their database design Wants to put data at risk just to make a poorly-written query run faster
- 43. Data Models are Complex …
- 44. Sample data Seller ID Book ID Book Name Price Seller Name AB ABC I Can Be VenusBarbie 9.99 AmazingBooks AB ABD Space Command Manual 10.00 AmazingBooks AB ADH Bookmark VenusBarbie 3.99 Amazing Books BD 3000001 I Can Be…VenusBarbie 6.00 Book Deals BD 3000015 Data Stories 27.00 BookDeals BD 4000200 Data Modeling Essentials 45.00 Book Deals BD 4000002 I Can Be…VenusBarbie 9.99 Book Deals DMC 110ABC Data Stories 28.00 Datamodel.com
- 45. Better… Seller ID Seller Name AB AmazingBooks BD Book Deals DMC Datamodel.com Seller ID Book ID ISBN/ UPC Price AB ABC 11111 9.99 AB ABD 22222 10.00 AB ADH 33333 3.99 BD X000001 11111 6.00 BD X000015 44444 17.00 BD 4000200 55555 45.00 BD 4000002 11111 9.99 DMC 110ABC 44444 28.00 ISBN/UPC Book Name 11111 I Can Be… VenusBarbie 22222 Space Command Manual 33333 Bookmark VenusBarbie 44444 Data Stories 55555 Data Modeling Essentials Bonus: Now we can keep data on Sellers and books we don’t have prices for yet.
- 46. Normalization & Data Volumes Less redundant data means less data over all Less redundant data means much faster DUI processing Less redundant data means faster backups, and faster recovery Less redundant data means smaller tables, indexes and databases
- 47. What Other Myths Do You Hear? Put them in the Q&A!
- 49. 7 Dangerous Myths 1. Data models are only for creating new databases 2. Data modeling is just documentation 3. DDL generated from data models is unusable 4. Data modeling isn’t needed – ORM scripts are good enough 5. I can code that myself faster & better than the data modeling tool can 6. All normalization just slows things down 7. Data modeling is just boxes and lines (the diagram is the model)
- 51. Thank you, you were great. Let’s do this again some time! Karen Lopez @datachick John Sterrett @johnsterrett